Official PDF , 38 pages - World bank documents

Public Disclosure Authorized Public Disclosure Authorized Public Disclosure Authorized Public Disclosure Authorized

21194 Education and Social Policy Department The World Bank July 1994 ESP Discussion Paper Series No. 33

\

>e

Public and Private Secondary Schools in Developing Countries: What are the differences and why do they persist?

Marlaine Lockheed and Emmanuel Jimenez

COPY FIllE The ESP Discussion Paper Series serves as an internal sysiem of record keeping, reference, and retrieval for the products of the Educazion and Social Policy Department's work program. The views expressed here are those of the authors and shoul not be auributed to the World Bank or its Board of Erecutive Directors or the counmties they represent.

Lockheedand Jimenez Page 2

Abstract

This paper presents the results of recent researchcomparingthe relative effectivenessand efficiencyof private and public secondaryschools in five developingcountries,which show a robust private school advantagein terms of studentachievementand unit costs. The paper exploresthe effect of school-based managementand autonomyand finds that principals of private schools were more likely to use their decisionmakingpower to improvethe conditionsfor learning in their schools. The studyhas important implicationsfor public educationpolicy, insofar as it suggestshow gains in publicschool efficiencymay be achievedby enablingpublic schoolsto adoptthe managementpracticesof private schools.


Public and Private SecondarySchoolsin DevelopingCountries: What are the differencesand why do they persist? MarlaineLockheedand EmmanuelJimenez

I. Introduction Most developingcountries provide public educationfree or at minimal cost to their citizens. Across the Third World, public schools enroll approximately90 percent of all primary and 70 percent of all secondarystudents. But becausefiscal constraintshavebeen increasingrecently, in manycountries the publicsector's abilityto expandthe provisionof free publiceducationhas beenlimited. In particular, this has createda seriousproblemfor the poorest countrieswhere the demandfor schoolingis projected to increase dramaticallyover the next decades. One way around this problem is to chargetuition fees for public educationservices, and many countrieshave adoptedthis policy. Another option is to rely on private schools to handle at least part of the expansion. Governmentscan encouragethis developmentby relaxing restrictions on private schools, by providing loans to and informationabout them and by restricting the number of places availablein publicschools. Some reports suggestthat, if governmentswere to adopt such policies, not only wouldmore resourcesfor educationbe generatedbut also standardsof efficiencyand qualitywould increase(WorldBank 1986). Becauseprivateschoolscompetefor studentsand are accountableto parents who pay the bills, they have an incentiveto adopt teachingpractices and to use staff and educational materialseffectivelyand economically.An additionalbenefitmightbe that, if publicschoolswere forced to competewith private schoolsfor students,they too might become more efficient. What is the scope for allowingprivate providers to play a larger role in handling educational expansionin developingcountries? This chapter addressesthis questionin the followingways.

Lockheedand Jimenez Page 4 o

We reviewbasic facts and figureson the public-privatebalancein educationand the policy debate surrounding it (Section 2). These facts show that, while the share of private enrollment in primary educationis uniformly small across world regions, there is much more variability in secondaryeducation,with some regions having25-50% of enrollmentin private schools.

O

We present the results of recent research comparing rigorously the relative effectivenessand efficiency of private and public secondary schools in five developing countries that are educationallydiverse -- Colombia, the Dominican Republic, the Philippines, Tanzania and Thailand. (Section3). These studies point to a robust private-schooladvantagein terms of achievement-testindicatorsand unit costs, after controllingfor studentbackgroundand selection.

o

We also peer into the 'black box" of private-publicschoolpracticesto try to explainthe observed differencesin the 5 countrycase studies, focusingon differencesin school inputs, processesand managementin two ways (Section4). First, we examine the existing data sets for relevant information;although each of the studies used data that were collectedfor other purposes they containedsimilar core information.' Second, we returned to each country and gathered more detailed informationfrom a small sample of public and private schools. We found that it may be possible to identifyprivate-schoolpracticesthat could boost performancein public schools.

In the final section of this chapter, we briefly discussthe role of public and private schools and explain why such a discussionis importantfor policy.

lI. Some Facts and FiLsures on the Public-PrivateRole in Education As pointed out by Roth (1987), "the concept of free compulsoryeducation,for which the state

'TDedata on Colombia and Tanzania were generated from a World Bank study of diversified education (see Psacharopoulos and Loxley 1985). The Philippines data were collected by the Ministry of Education as part of its Household and School Matching Survey. The Thailand data were obtained from the Second International Mathematics Study conducted by the lnternational Association for the Evaluation of Education Achievement(IEA) (see Robitaille and Garden, 1989). The Dominican Repubtic data came from a survey modeled afterthe IEA (Luna and Gonzalez, 1986).

Lockheedand Jimenez Page 5 should be responsible,originatedin Europe and North America but was not widelypromoted until the nineteenthcentury. It is thus of comparativelyrecent origin. Private educationhas had a much longer history..." (p. 16). What is the situationtoday? Recently,UNESCOhas againbegun to publish data on the role of private schoolsin education. Thesedata are largelyself-reportedby countriesand reflectthe huge variety in types and financing of private schools.2

In particular, while most schools in the database are

consideredto be private on the basisof ownership,the degreeof governmentregulationand subsidization varies widely. Nevertheless,Table 1 offers some roughorders of magnitudeon the breakdownof public versus private schools in the 1980s. There are several findings to highlight. One is that primary and secondary enrollment is predominantlyin public. While the overall numbersobscurethe data of individualcountries,nearly all the (unweighted)regionalaveragesfor both primaryand secondaryeducationshow a public sector share higher than 60 percent. Another findingis that there is significantdisparityacross levels. The share of public education at primary level is substantiallyhigher than that at secondarylevel. In Asia, for example,in 1975the averageshare of public enrollmentwas sixteenpercentagepoints higherin primary than in secondaryeducation. Also, there is a large variance across regional groupings, particularlyat the secondarylevel. At the primary level, the share of public educationhas remainedat 80-90%for all regions other thanAfrica since 1965. At the secondarylevel, however,public educationshares grew until 1980; more recent anecdotalevidencesuggestthey have declinedsincethen.

2 Definition of Private Education: "The International Classification of Education defines private education as that provided in institutions managed by private persons. This definition covers a wide variety of situations. Some private institutions are whoUy funded by the State, others are state-aided to a wide variety of degrees while others again receive no state aid at all. In any one country, the situation may vary over time or according to level or type of education". (OECD 1990, p.40).

Lockheed and Jimenez Page 6 Tables 1: Percentage Share of Public in Primary and Secondary Education

1965

OECD Asia East Africa West Africa Middle East & North Africa Latin America & Caribbean Low-Income Middle-Income

87 86 47 74 92 86 75 80

Primary 1975

88 87 57 82 94 87 81 85

1985

1965

Secondary 1975

89 88 80 84 98 88 95 84

n.a. 56 64 59 82 61 n.a. n.a.

n.a. 71 55 70 92 67 n.a. n.a.

1980

n.a. 78 52 72 91 75 n.a. n.a.

Note: Figures for low-income and middle-income countries are incorporated from Table A-10 in Lockheed and Verspoor (1991). All other 1965 and 1975 primary education data and all secondary education figures are from Tan, as quoted in Roth (1987) Table 2-1. OECD data and 1985 data for Asia, East Africa, West Africa, Middle East and North Africa and Latin America and Caribbean are from the UNESCO data base official enrollment series. The figures do not reflect averages of constant cases.

The figures in Table I masktremendousdifferencesamongschoolsacross (andwithin)countries. In particular, to discern how the role of the private sector affects educationaloutcomes, it is important to have informationabout at least three critical characteristics: financing (the extent to which private schools are subsidized),regulation(howstrict is governmentcontrol over the way the schools are run) and ownership (whetherthe private schoolsare sectarian,religiousor for profit). There is very little systematicallygathered evidence on these characteristics. The most comprehensiveis probablya recent reviewby James (1991)of the experiencesof some 35-50developing and developedcountries. Her study indicatesa large variancein experienceamongcountries. In general, private schools are heavily subsidized(to the extent of 80 percent or more of costs) in those countries (most of which are in Europe) where teacher salaries are paid by the state.3 These are also the most heavily regulatedschools. A wide variety of countriesprovide indirect (through tax breaks) or partial

'Somedevelopedcountries, such as Lesotho,Togo and Chile, and the state of Kerala in India also support their private schoolsin this way.

Lockheed and Jimenez Page 7 subsidies(ess than 25 percent of costs)to privateschools. Theseinclude countriesas diverse as Japan, the U.S., the U.K., Indonesia,Kenya and severalLatin Americancountries. These governmentstend to have only a moderate amount of control over individual schools. On the other hand, some governmentswhichoffer neitherindirectnor partialsupport,still attemptto regulatetheirprivate schools. The quality of private schoolsand the type of subjectsthey emphasizealso varies dependingon their role vis-a-vis the public sector. The next sectionanalyzesthis importantrelationship. Public and Private Schools: The Policy Debate It is generallyacceptedthat educationis a responsibilitythat is shared betweenthe family (or student) and the state. The role of the state is importantfor a number of reasons. First, when one individual's consumptionof a good affects others (an externality),the individualmust be inducedto considerthe social as well as the private costs and benefitsof his behavior. It is often arguedthat this is relevant in many aspects of education, particularly at the primary level. Second, government interventionmay be necessary because financialmarkets are too limited to allow studentsto borrow enoughto covertheir current costs on the basis of their likely future earnings. Third, if makinghuman resource investmentsis acceptedas a principalstrategyfor alleviatingpovertythen this also impliesthe need for governmentaction (WorldBank, 1990). Theseobjectivesprovidean economicjustificationfor the most commonfinancingpracticefound in most countries -- the reliance on general revenuesto finance public education. But what is the role for private educationin this setting? First, private educationmust fill the inevitablegaps in public educationprovision. A lack of firm political support can sometimes limit the degree to which governments can assist public schools to develop (see James and Birdsall, 1991; Birdsall and James, 1992). Even when they are politically motivatedto do so, as in many developingcountries,governmentsare often under such severe financial constraintsthat they cannot afford to financeeven a high return activitysuch as education(WorldBank,

Lockheed and Jimenez Page 8 1986). The result is an excess demand for public places. James characterizestwo other different motivationsfor the establishmentof private schools- a differentiateddemandarisingfrom a deep-seated religious or linguistic diversity and an offer on the part of an entrepreneur or organization,.often religious, to start the schoolson a nonprofitbasis. She hypothesizesthat, in manydevelopingcountries, it is excessdemandthat promptsthe developmentof privateeducation,whereasit is differentiateddemand that has the same effect in developedcountries. This situationmay change as countries developand differentiateddemand becomes a bigger motivator of private education (James 1988, 1989a, 1986b, 1986c). A secondrole thatprivate educationcanhave is in fosteringgreater efficiencyby requiringpublic schoolsto competefor students. Althoughthere are manytypes of private schools, analystshave argued that they possessgeneral characteristicsthat differentiatethem from publicschools. These characteristics includea greater flexibilityin the way theyoperate and in the way they are funded, a direct accountability to those who use their servicesand a greater tendencyfor those in chargeof individualschoolsto make critical educational decisions (Coleman, Hoffer and Kilgore, 1982). It is often argued that these characteristicsenable private schools to provide educationmore effectively, in other words, to provide the type and quality of educationstudents and their parents demand. Moreover, even when the quality of the provision is the same, private schoolshave the incentiveto provide an educationat a lower cost than do their public school counterparts. Third, private schools can serve as a laboratory for alternative models of school-level management,which, if effective,could be adoptedby public schools. Theoretically,private schoolsare free of the bureaucratic constraintsthat encumberpublic schools, and are able to control many more decisionsat the school level. It is arguedthat the greater school-levelautonomyof private schoolscould be employedby public schools to boost achievement. Not all analystsagree that public schools would be able to emulateprivate schools, because of the differencesin their sources of support. They reason

Lockheed and Jimenez Page 9 that private schools emphasizelearningbecauseparental fundingmakes private schools accountableto parents and to their demandsfor effectiveinstruction. By comparison,public schoolspay less attention to learningbecausethey needto balanceprofessionalaccountabilityto parentsagainstfiscal accountability to the centralizedfunding source. If correct,thesetheorieshavesubstantialpolicyimplicationsfor developingcountries. At present, private schoolsplay only a peripheralrole as a conduitfor educationalexpansion. Can this be changed? Shouldprivate schoolsbe deregulated? Shouldtheybe subsidized?Some analystshave arguedthat these schoolsare only for the elite. Othershave said that private schoolsshouldnot be allowedto proliferate because they will provide low quality educationat high cost to gullibleparents and students. Like all good theories, these lead to the followingquestionswe explore in the rest of this chapter: o

Are private schoolsmore effectivethan public schools?

o

Are private schoolsmore efficientthan public schools?

o

Is the greater effectivenessand efficiencyof privateschoolsdue to differencesin schoollevel management?

m. Effectivenessand Efficiency' What is the empiricalevidenceregardingthe relative efficiencyof private and public schools? In the United States, the provocativeColeman,Hoffer, and Kilgore (1982) report had concludedthat attendingprivate schools improvedthe performanceof students as measured by standardizedtests of verbal and mathematicalskills. Althoughthere are outstandingquestions of selectivitybias and the magnitudeof effect(see, for example,AmericanSociologicalAssociation1982;Murnane1985;Murnane, Newsteadand Olsen 1985), the conclusionthat the average studentdoes better in private than in public

'This section draws heavily from Jimenez, Lockheed and Paqueo (1991).

Lockheed and Jimenez Page 10 schoolsis widespread(Hanushek1990).5 For developingcountries,the recent evidenceis muchmore recent. We present here the results of a researchproject we managedwhich comparedrigorouslythe relative effectivenessand efficiencyof private and public secondary schools in five developing countries that are educationallydiverse -Colombia,the DominicanRepublic, the Philippines,Tanzaniaand Thailand. We focusedon secondary 6 schools becausethis is the level at which private participationis most significant.

All the case studies address the followingquestion: Would a high school student, selectedat random from the general studentpopulation,do better in a public or private school? In the absenceof experimentaldata, the studies compare students' performanceon standardizedtests in a cross-sectionof public and private schools,with studentbackground,motivationand innateability and prior performance controlledthrough the use of various statisticaltechniques. These techniques purge the influenceof backgroundfactors from the achievementscores. They also ensure that there is enough overlap in the distributionof characteristicsof the studentsso thatthe subsamplesare truly comparable. Other statistical techniquesare then used to control for possible selectionbias. Because it is difficultto measure manynonschoolor family backgroundeffects (for instance, innate ability), we supplementthe cross-sectionalstudies with studiesusing panel data that comparethe differencesin the achievementof public and private school students over two time periods. Nonschool effects that do not changeover that time are netted out. In the studies of the Dominican Republic and Thailand, changesin achievementacross two time periods are used rather than the level of achievement in a given time period. As far as we know, this type of value added analysis of private schools has

'The magnitude of this advantage has been disputed (Wllliams (1985)). Levin (1987) claims that the estimated gain in studcnt achievement, particularly in longitudinal data sets, would have very little effect on cnhancing the chances for college admission in the U.S. or on wages. Other recent studies, such as Roth (1987), James (1987) and Samoff (1987), look at the private sector's role in providing educationin developing countries but do not compare costs or achievement in private and public schools. Also see Jimenez and Lockheed (1991) for studies that examine public and private schooling issues more broadly.

Lockheedand Jimenez Page 11 previouslyonly been done on a data set from a developedcountry(for data on U.S. high schools, see Coleman,Hoffer, and Kilgore1982; Lee and Bryk 1988; and Hanushek1986). Table 2 summarizesthe salient featuresof the case studies. To examine the comparative efficiency of public and private schools, we also compare differencesin achievementwith differencesin costs. Althoughthe results described in this chapter are uniquein that they combineboth effectivenessand cost comparisons,the few studiesthat applyonly parts of the methodologycorroborateour results (Lunaand Gonzalez1986;Psacharopoulos1987; Tsang and Taoklam 1990.) Table2: Summary of Studies Year data collected

Colombia

1981

Numberof Students Schools

1,471

35

Grade

11

Indicatorof achievement

Averagescores on math and verbal tests

76

1982-83

2,472

Philippines

1983

446

Tanzania

1981

1,025

13

4,030

99

DominicanRepublic

Thailand

1981-82

Data base

Special survey

8

Mathematics test

National school survey

7-10

Mathematics test English test Pilipinotest

National household survey

11

Average scores on math and verbal tests

Special survey

8

Mathematics test

National school survey

Relative Effectiveness of Public and Private Schools Do private schools provide a better education than public schools? A principal finding in all five

Lockheed and Jimenez Page 12 countriesis that, givenstudentbackground,studentsin private schoolsgenerallyoutperformtheir public counterpartson standardizedmathematicsand/orlanguagetests. We reach this conclusionafter carrying out the followinganalyses. The predictedscores in each typeof school are obtainedfrom the regression equationsrelatingbackgroundto achievement,as evaluatedat the level of backgroundcharacteristicsof the averagepublic school student. This holdsconstantfor the effects of background. Table 3 shows the ratio of a student's predicted score in a private school to his or her score in a public school. For example,in Colombia, a studentwith the backgroundof the averagepublic school studentwould score 1.13 times (13 percent) better in a private schoolthat in a public school. This ratio varies considerably across countriesbut is consistentlygreater than one for all subsamplesand achievementtests (with the possibleexceptionof mathematicsachievementin the Philippines,wherethe differencesare insignificant). In terms of standarddeviationunits, or "effectsize", theprivate schooladvantageis large and meaningful in all cases, ranging from one-halfto two standarddeviations(Cohen 1969). These differences in the effect size associatedwith private schools cannot be dismissedas trivial (Levin 1987).

Lockheed and Jimenez Page 13 Table3: The PrivateSchoolAdvantage: PredictedTest Scorein Private Schoolsas a Multiple of PredictedTest Score in PublicSchools and in StandardDeviation Units

Indicatorof Achievement

Relative Advantage

Averageipath and verbal Mathematics(0-type) Mathematics(F-type)

1.13 1.31 1.47

0.55 0.89 2.16

Philippines

Mathematics

1.00

-0.09

Tanzania Thailandv'

English language Filipino language Averagemath and verbal Mathematics

1.18 1.02 1.16 2.36

0.33 0.25 0.97 1.69

Country Colombia DominicanRepublic!'

Effect Size

Note: The tableshowsthe proportional gainin achievement scoreif a randomlyselectedstudent,withthe characteristics of the averagepublicschoolstudent,attendsa private ratherthan a publicschool,holdingconstantthat student's background. RepublicandThailand,thetest scorebeforethe schoolyearbeganwasincludedas a regressorin a! For the Dominican the equationexplainingachievement at the endof the year. The phrase "given student background" is critical here. It is generally not valid to infer differences among types of schools based simply on comparing achievement on standardized tests because students' backgrounds vary so much between types of schools, reflecting differences in choice. Because the private schools in our case study countries charge tuition while the public schools are almost free, the most important factors in the household decision are income (or income related variables such as parents' education and occupations) and the relative cost of schooling. According to Table 4, in Colombia and the Philippines, average income indicators for students in private schools are about twice as high as are those for students in public schools.

Interestingly, in Tanzania, this difference is much lower, which

suggests that public schools attract students from higher income families. These findings are corroborated by data showing that private school students in Tanzania tended to come from families where the father had a white collar job and the mother had some education. In both Thailand and the Dominican Republic, private school students come from families with more educated mothers and with fathers who were employed in white collar occupations. The range in income, however, is only slightly higher for

14

Lockheedand JimenezPage 14

private than for public school students in Colombiaand lower in Tanzaniaand the Philippines, which suggestsa substantialoverlap in the income categoriesof the public and private school samples.7

Table 4: BackgroundIndicatorsfor Private School Studentsas a Multiplefor Public School Students

Colombia

Dominican Rep. 0-type F-type

Philippines

Tanzania

Thailand

Income (of household head or father)

1.94

2.07

1.20

Coefficient of variation of income

1.24

0.72

0.3

Mother's education (percentage beyond primary)

1.87

1.62

2.21

1.23

1.27

1.61

Father's occupation (percentage white collar)

1.09

1.69

2.52

1.50

1.94

Percentage male

1.04

1.29

1.78

1.07

0.91

iNote:

a/

0.98

The table shows the extent to which an indicator for private school students exceeds that for public school students. For example, in Colombia, the average household head income of students in private school is 1.94 times (almost twice) that of students in public school. A figure close to one implies that an indicator for private school students is equal to that for publie school students. F-type schools are authorized to give Ministry of Education examinations. 0-type schools are not so authorized.

After we hold the backgroundeffects constantby measuringachievementat the average characteristics of public or private school students, we find that the advantageconferredby private schools is greater for the two countries with the best controls for student background-the Dominican Republic and Thbailand.The data sets for these studentscontainedtest scores measuredat the beginningand at the end of the school year, and the ratios measure changethe achievementover the course of the academicyear (with controls for possible sampleselectionbias). Do theseresults hold for studentsfrom differentsocioeconomicgroups? Qualitatively,the answer is yes. The private school advantagepersists even when the computationsin Table 3 hold constantthe backgroundof the average private school student, whosestatus is higher than that of the average public school student. The Philippinesstudy is the only one that lookedat the sensitivityof the private-public

'In all cases, the analyses controlled for selection bisa - choice - through statiatical methods which are fully explained in Jimenez, Lockheed and Wattanawaha(1988).

15

Lockheed and Jimenez Page 15

differentialto a wider range of socioeconomicindicators,variations in socioeconomicstatus, within a reasonable range, did not reverse the private school effect. But the magnitudeof the private school advantagesubstantiallydecreaseswith lower socioeconomicstatus. This is consistentwith the fact that the more elite private schoolsin the Philippinestend to emphasizethe developmentof Englishlanguage skills and that advantagedchildrenhave more exposureto English and better accessto Englishlanguage media. Among children who speak Filipino, on the other hand, there is no relationship between socioeconomicstatus and attendingprivate school. And in mathematics,there is a virtual tie between public and private schools. Can peer group characteristicsaffect student achievement? In the Dominican Republic and Thailand, the only two countries for which data were available, peer group effects (the academic backgroundor social class of studentsin each school)were very important. Relative efficiencyof publicand privateschools. What about efficiency? Preliminarycalculationsbased on school expendituredata indicatethat, on average, the unit costs for private schoolsare lower than those for public schools (Table 5, column 1)). Thus for the same unit cost, private schoolsprovideas much as three times more learning as the public schoolsdo (Table 5, column2). Conversely,the same amountof learning in private schoolscan cost as little as 15percent of its cost in public schools(Table 5, column3). These results indicatethat private schools are more efficient than public schools, at least in-secondary schools in the sample countries. But ther,eare some importantcaveats. First, the orders of magnitudeare rough. The cost estimatesfor Colombiaand Tanzaniaare not precisebecausea numberof private schoolsdid not provide the necessaryinformation. Second, in the Philippines,we used the average cost for a nationwidesample of schools (basedon World Bank sector work), rather than the actual cost of the schools in the study. By comparison, in the DominicanRepublic and Thailand, we had school-by-schoolcost data for the sample. Third, the cost figures generallydo not includeeducationalexpenditures,such as books,supplies

16


and uniforms,that are not paid to schools. We do not, however,expectthese data to causethe qualitative resultsto change significantly. Generally,nonschooleducationalexpenditures,such as books, supplies and uniforms, are higher in private schools. And interviewsin the countriesstudied reveal that even religiousprivate schoolstend to use lay teachersrather thanpriests and nuns. Moreover, subsistenceand other nonsalarypersonnelcosts are coveredin the cost data. Finally, there is considerablevariabilitywithineach schooltype, as noted in the data on private schools in the Dominican Republic. Philippinepublic schools (say, those that are primarily locally funded)have lower unit costs than some types of private schools (the elite schools). Unfortunately,the survey datadid not distinguishstudentachievementamongtypes of public schools.It wouldbe interesting in any subsequentanalysisto explorethis comparison.


17

Table5: RelativeAverage Cost and Efficiencyof Publicand Private Schools

Country

v' l

(1) Ratio of private cost to public cost

(2) Ratio of relative effectiveness to coste'

(3) Ratio of relative cost to effectivenesse

Colombia

0.69

1.64

.61

DominicanRepublic 0-type F-type

0.65 1.46

2.02 1.0k

.50 .99

Philippinese Math English Filipino

0.83 0.83 0.83

1.40 3.17 1.20

.83 .70 .82

Tanzania

0.69

1.71

.59

Thailand

0.39

2.62

.15

Figures from Table2 dividedby column1 of Table 3. 1Column 1 of Table 4 dividedby figuresfrom Table2. Public cost estimates,weightedaverage of nationaland local costs. Costs are assumed to be the same for all three subjectsand are based on World Bankestimates.

IV. Lookinginto the Black Box Why are private schoolsmore effectivethan publicschools in boostingstudentachievement?This section explores the effects of school-levelresources and inputs and goes "inside" public and private schoolsto examinehow they are organizedand managed. Here, we raise the questionof whether it is possible for public schools to reorganizethemselvesalonglines developedby private schools. One reason that private schools may be more effectivethan public ones is that they have more resourcesto bring to the classroom:better educatedteachers,more instructionalmaterials,a larger stock of institutional resources such as libraries, laboratories, or subject rooms. Another explanation is that

they are more likely to choosethe mix of inputs that acceleratesstudentlearning, economizingon those

18


inputs that have little impacton studentlearning. We have seen that per-studentexpendituresin private schoolsare lower, not higher, than those in public schools,which suggeststhat the explanationfor their greater effectivenessdoes not lie in the greater abundanceof resources in general. We now comparethe types of resources availablein public and private schools, to see what, if any, differencesemerge in the mix of inputs each choose. One reason that private schools may be more effectivethan public ones is that they have more resourcesto bring to the classroom:better educatedteachers,more instructionalmaterials,a larger stock of institutionalresources such as libraries, laboratories,or subject rooms. Another explanationis that they are more likely to choose the mix of inputs that acceleratesstudentlearning, economizingon those inputs that have little impact on student learning. We have seen in previous chapters that per-student expendituresin private schools are lower, not higher, than those in public schools, which suggeststhat the explanationfor their greater effectivenessdoes not lie in the greater abundance of resources in general. The case studies were able to supplysome informationregardingdifferencesin resources between public and private schools. We have already notedthat the overalllevel of resourcesavailablein private schools is lower than that available in public schools. In addition, althoughit is not possible to infer causalityfrom Table 6, it shows some interestingcomparisons. In all countries,private schools tend to choose slightly higher student-teacherratios and to use the savings to purchase other inputs. For example, in Thailand,private schoolsmake more efficientuse of teachers by recruitingcandidateswith slightly lower qualifications, giving them more in-service training and by promoting better teaching processes (homework, tests and orderly classrooms). In the Dominican Republic, the most striking difference is that students in private schoolshave better access to textbooks.

19

Lockheed and JimenezPage 19

Table 6: Average Private SchoolInput and ManagementCharacteristicsas a Multiple of Average Public School Characteristics

Variable

Colombia

Dominican Republic Tanzania 0-type F-type

Thailand

MiniSurvey

Input Mix Teacher salary Student-teacherratio/classsize Teacher's years of education Minutesspent on maintaining class order Proportionof studentswith textbooks Proportionof teachers Qualifiedto teach math in student's school With in-servicetraining Teachingenrichedmath class Total resources Total instructionalmaterials Management Total school-levelautonomy Instructionaltime

0.52 0.85 -

1.00 .95 .38 3.11

1.00 1.02 1.21 3.50

1.15 1.07

-

-

-

-

-

-

1.05

1.10

-

-

-

-

1.24 -

1.06

-

-

-

-

0.17 2.29 1.54 -

0.87

-

-

-

-

1.61 1.08

-

0.97 1.11

Private schoolsmay also be more effectiveand efficientbecauseof their internalmanagement. Hannaway(1991)argues that schoolsand school systemsorganizethemselvesto enhancethe objectives of their resource providers. In the case of private schools, these are parents; in the case of public schools,theyare local, regional and nationalauthorities. The result is that private schools adopt an organizationalstructurethat is flexiblewith respect to the needs of individualstudentsand that is aligned with the professional interests of school level personnel. By*comparison, public schools adopt an organizationalstructurethat servesto promoteinternalcompliancewith the multipleobjectivesof not only individualparentsand studentsbut also the wider society. Hannawayalso argues that so long as public schools receivetheir financing directlyfrom "central" sources, they will not be able to adoptthe more flexibleorganizationalstructure chosenby private schools in response to parental sources of funding.

20


Efforts at decentralizationhave attemptedto shift the responsibilityfor resource provision to local levels, in an attemptto encouragegreater school-levelresponsivenessto local needs. However, in many cases, shiftingfinancialresponsibilityhas been viewed as a means of relieving a burden on the centralbudget, not as a means of providinggreater autonomyto local schools. The result has been that decentralizationhas neither suppliedmore materialresourcesat the schoollevelnor enhancedschool-level autonomyover decisionmaking (Lockheedand Zhao, 1992). School-levelautonomyis a key differencebetweenpublic and private schools. In her research on Catholicand public schools in the United States, Hannawayfound that school autonomywas higher for Catholic schools than for public schools, even with a number of adjustments for student characteristics,organizationalcontext, and principalcharacteristics. She notes that "even if we changed public schoolsin terms of their clients, their organizationaland politicalcontexts, and the characteristics of their principals, these schools would still not be managed the way private schools are managed" (Hannaway, 1991, p. 473). The reason is that as long as public schools are accountableto central bureaucracies,they will be organizedin a managementstructurethat limits school-levelautonomy. To answer the questionregardingthe causeof the observed greater effectivenessand efficiency of private school relative to publicones, we undertooka "mini-survey"of public and private schools in the five case-study countries. This survey was undertaken from a distance. We invited a senior researcher in each countryto gather systematicdata abouta variety of institutionalpracticesin public and private schools, using a survey instrumentwe provided. In each country,the researcher was asked to identifythree schoolsin each of the followingcategories: private elite, private nonelite, public elite and public noneliteschools. This made a total of 12 schoolsper country. All of the schools were visited by the researcheror a representativeof the researcher,who interviewedthe headmasteror principalteacher. Althoughtwelve surveyswere returnedfrom each country, for a total of 60 schools, all types of schools were not representedequally. We receivedreturns from 14 private elite schools, 17 private nonelite

21


schools, 13 public elite schoolsand 16 public noneliteschools. The results from the mini-surveyconfirmedmuch of what we have observed previouslywith respect to student selectivityand resourcesimilaritiesamongschools in the two sectors. The private schools in this survey were not systematicallymore advantagedthan the public schoolsin terms of the material resourcesavailableto them, althoughtheir studentsmay have been more capablethan those in the publicschools. The mini-surveyalso confirmswhat we had suspected,that privateschools are more advantagedin terms of their abilityto managethemselvesand make educationaldecisionsat the level of the school site. Resources In manyrespects,the publicand privateschoolsin the mini-samplewere similar (Table7). The majority of both the private and the public schools were coeducational(81 percent and 86 percent respectively)day schools,with only 20 percenthavingany type of boardingfacility. Approximatelythe same proportion(45 percent) were classifiedas "elite" schools(a consequenceof the instructionsissued to the researcher). However, despite these similarities,there were significantdifferencesbetweenthe public and privateschools in our mini-survey. Publicschoolsadmittedmore applicants(36percent) than did private schools(23 percent) and were less likely than public schools to admit studentson the basis of test performance(55 percent and 65 percent respectively). Second, a higher proportion of teachers in publicschools were "fully qualified"than in private schools. Third, the public schoolswere larger, enrolling more students and employingmore full time teachers;possibly as a result, the public schools were more likely to operate on two shifts. Fourth, the public schoolstaughtfewer differentgrades than the private school (5.8 grades and 8.5 grades respectively). While public schoolsdiffered from private schoolsin terms of the numbers of students and teachersthey had and in the numberof grades offered, they had only one more administrator,on average, than private schools had. The picture that emerges here is of administratorsin private schools managingsomewhatmore able students in physicallysmall

22


but educationallycomplexorganizations,and administratorsin public schools managingsomewhatless able students in physicallylarge but educationallysimple organizations.

Table 7: General Characteristicsof Public and Private Schoolsin Mini-sample,Colombia,Dominican Republic,Philippines,Tanzaniaand Thailand1990 Characteristics Percent coeducational Percentresidential Percent elite Number of shifts Number of students Number of grades Number of full time teachers Percent fully qualifiedteachers Number of administrators

Private 80.6 19.3 45.2 1.4 1113.1 8.5 43.5 89.7 7.0

Public 86.2 20.7 44.8 2.0 1917.0 5.8 85.1 104.0 8.0

Physically,the public and private schoolsin the mini-surveydifferedvery little and both sectors appeared relatively advantaged(Table 8). In particular, the conditions for teaching did not differ significantly between public and private schools. Most students were accommodatedin regular classrooms; most classrooms had seats and desks for both students and the teacher. Virtually all classroomshad blackboards. Both public and private schools reported having an average of 11 of 13 important physical resources - file cabinets, telephone, typewriter, television, computer, radio, duplicatingmachine, school library, science laboratory,subjectrooms, storage facilitiesfor books, staff room, officefor school head and a kitchen. The only differencebetweenpublic and private schools was that more privateschools than public schoolsreportedhavinga school library (93percent and 79 percent respectively),whereas more public schools than private schools reported havinga copyingmachine (85 percent and 61 percent respectively).

23


The physical plants of public and private schools were also similar with respect to selected "modern"amenities. All private schoolsand 83 percent of public schools reportedhaving water piped to the schools, and aboutthree quartersof both public and privateschoolsreportedhavingelectricitythat worked regularly. Most schools reported having flushabletoilets for students (74 percent of private schools and 83 percent of public schools). Both public and private schools attemptedto protect the buildingfrom vandalism,althoughmore publicschoolsthan private schoolsemployedsomeoneto guard the school at night, on weekendsand over holidays(90 percent and 77 percent respectively). Private schools reported greater concern with on-going maintenance;three quarters of the private schools employeda regularmaintenanceman, whereasonly 43 percentof publicschoolsemployedsuch a person. From this descriptionit seems that the schoolsin the mini-samplemay not represent typical secondary schoolsin developingcountries. Reflectingthe similaritiesin the overallphysicalplant and resourcesavailableat the schoollevel, the number of instructional materials available to teachers in public and private schools were approximatelythe same (Table 9). Teachersin both typesof schoolshad aboutfive out of a list of eight importantinstructionalmaterials- chalk, writing implements,paper, instructionalguides, illustrations, sciencekits, textbooksand dictionaries. However,more private schoolteachers (66 percent) thanpublic school teachers (52 percent)had a storage cupboardin which to store these supplies. Studentsin private schools appeared slightly more advantagedthan those in public schools. While most students in both public and private schoolshad writing implementsand paper, private school studentswere more likely to have dictionaries(63 percent) and a completeset of textbooks (70 percent) than students in public schools(40 percent and 66 percentrespectively). The differencescouldreflectthe differencesin student selectivitybetween public and private schools or could reflect investmentchoices on the part of the schools.

24


Table 8: Physical Characteristicsand Facilities of Public and Private Schools in Mini-sample, Colombia, DominicanRepublic, Philippines,Tanzaniaand Thailand1990 Characteristics

Private

Student-teacherratio Numberof classrooms Numberof resources V Percent studentsthat have class outside Percent studentswith seats & desks Percent classroomswith teacher desk Percent classroomswith blackboard Percent with health facility Percent with regular electricity Percent with piped water Percentwith flushabletoilets Percent with buildingguard Percent with maintenanceperson

24.2 27.0 11.2 0.9 99.7 90.4 100.0 50.0 79.3 100.0 74.2 77.4 74.2

Public 22.1 29.6 11.5 5.2 95.7 81.8 99.8 50.0 75.9 82.8 82.8 89.7 42.9

e Filecabinets,telephone,typewriter,television,computer,radio,duplicatingmachine,schoollibrary,sciencelaboratory, subjectrooms,storagefacilitiesfor books,staffroom,officefor schoolhead,kitchen.

Officialinstructionaltime was also relativelysimilar for public and private schools (Iable 10). Both types of schools reported having a school year of approximately200 days, a school day of seven or eight periods, and periods lasting about50 minutes. However, public schoolswere closedfor nearly four times as many school days (fifteen)as private schools (four). The overall resources availablein public and private schoolsin the mini-surveywere remarkably similar. However, where differencesemerged, they suggest a decisionon the part of private schoolsto invest in resources more closely aligned with instructional goals: libraries, dictionaries, textbooks, classroomstorage areas for instructionalmaterials, instructionaltime. By comparison, public schools tended to invest in personnel: teachers and buildingguards, and to pay for teacher qualifications.


25

Table9: InstructionalMaterialsin Publicand PrivateSchoolsin the Mini-sample,Colombia,DominicanRepublic, Philippines,Tanzaniaand Thailand1990

Public

Characteristics

Private

Number of materials a/ Percent of teacher with storage cupboards Percent of students with pen/pencils Percent of students with paper/notebooks Percent of students with complete sets of texts Percent of students with dictionaries

5.0 65.7 96.7 96.6

4.5 51.9 99.1 96.4

70.0 63.2

66.3 40.1

-/

Chalk, writing implements,paper, instructionalguides, illustrations,sciencekits, textbooksand dictionaries.

Management Public and private schools differ significantly in terms of their management organization.

In most developing countries public schools are financed and managed by the central government. Teachers are hired and deployedby a central agency, curriculumis set nationally, and admissionto secondaryschool is often controlledby nationalexaminationswith students placed in schools through centralagencies. As a result, neitherthe local communitynor the schoolprincipalexercisesmuchcontrol over key decisions. Unlike centrallycontrolledpublic schools, private schools in both developedand developingcountriesexercise managerialcontrolover a wide range of decision. For example,research has foundthat in U.S. Catholicprivateschools,principals,teachersandparents havesignificantlygreater control over decisions about the curriculum,instructionalmethods, allocating funds, hiring teachers, dismissingteachers, and disciplinepoliciesthan do theircounterpartsin public schools(Hannaway1991). Hannawayconcludesthat "there is somethingabout public educationalinstitutionsthat restricts their adaptationto local conditions"(Hannaway1991,p. 122). Similardifferencesin patternsof controlwere found in the Philippines(Lockheedand Zhao, 1992).

26


Table10: InstructionalTime in Public and Private Schoolsin the Mini-sample, Colombia,DominicanRepublic, Philippines,Tanzaniaand Thailand1990

Characteristics

Numberof daysin schoolyear Numberof periodsin schoolday Lengthof instructional periodin minutes Numberof daysschoolwasclosed

Private

Public

202.0 7.2

203.0 8.0

47.0 4.3

48.0 15.1

In the mini-surveylarge differences betweenthe public and private schools emerged in two regards: (a) the degree of influenceexercisedby the principalover school level decisionmakingand (b) the importanceapparentlyplaced on academicachievement. Private school principals reported more influenceover school level decisionsand greater attentionto matters of teachingand learning. We listed 13 areas of school level decisions - selectingteachers, selecting nonteacherstaff, dismissingschoolpersonnel, selectingteachers for in-service,evaluatingteacherperformance, adapting the curriculum,establishingstandardsfor studentpromotion,improvinginstructionalpractice, choosing textbooks,purchasingequipment,establishinghomeworkpolicies, selectingstudents,settingand spending school fees. Then, we asked principals to indicate which of five groups (head office or Ministry of Education,schoolboard, headmasteror principals,teachers,parentsor PTA) exercisedthe most influence over each area. Principals of private schools were more influentialover these areas than principals of public schoolsin two respects(Table 11). First, principalsof private schoolsinfluencedmore areas of decision making than principals of public schools (5.5 areas versus 3.4 areas respectively). Second, principals of private schools influencedmore of thesedecisionsthandid any othergroup, includingthe headoffice. Principals of private schools influenced over five times as many areas (5.5 areas and 1.1 areas, respectively)as did the head office. By comparison,principals of public schools influencedfewer of


27

these school level decisionsthan did the head office (3.4 areas and 3.9 areas respectively). As a consequence,the head office or Ministry of Educationinfluencednearly four times as many decisions in public schoolsas they influencedin private schools. Table11: InfluenceOver 13 SchoollevelDecisionsi1 in Publicand PrivateSchoolsin the Mini-sample,Colombia, DominicanRepublic, PhilippinesTanzaniaand Thailand1990

Characteristics

Private

Public 3.92

Head/Office Ministry of Education

1.11

School Board

1.88

1.44

Headmaster/Principal Teachers Parents/PTA

5.46 3.23 0.00

3.40 3.56 0.16

S'

Decisionsover: selectingteachers, selecting nonteacherstaff, dismissingschoolpersonnel, selecting teachersfor inservice,evaluatingteacherperformance,adaptingthe curriculum,establishingstandardsfor studentpromotion,improving instructionalpractice, choosingtextbooks,purchasingequipment,establishinghomeworkpolicies, selecting students, setting school fees, spendingschoolfees.

We also looked at each specificdecisionmakingarea. Principalswere the most influential group in at least 40 percent of the private schools for eight areas- selectingteachers for in-service, purchasingequipment,evaluatingteacher performance,selectingnonteachingstaff, selectingstudents, dismissingschoolpersonnel, selectingteachers and spendingschoolfees. By comparison,in the public schools, principals were most influentialover only tw areas - evaluatingteacher performance and selectingteachers for in-service(Table 12). Another explanationfor the higher achievementin Catholicprivate schoolsversus public schools in the United States is that they place greater emphasison engagementin academicactivities, includinghigher rates of enrollmentin academiccourses. This, in turn, translatesinto such differences in student behavior as spending more time on homework (Colemen, Hoffer and Kilgore 1982). In developing countries, curricula are typicallyset nationally, and students have little choice over course selection. However, differencesin the emphasisplaced on academicachievementmay vary between schools, and this may translate into differencesbetweenpublic and private schools in the level of effort

28


spent by students on academicactivities. In the mini-survey,we found thatprivateschoolprincipalsnot only had significantinfluence over what occurred in their schools, but also establisheda school climate that promoted learning and rewardedthose that contributedto its success. The mini-surveyshowedthat private schoolsemphasized teaching and learning more than public schoolsdid and that they providedrewards that were contingent on good performance (Table 13). Both in-servicetraining and regular staff meetings were more likely to be designedto strengthenteachingmethodsthan were similar activitiesin public schools. Virtually all private schoolsofferedmonetary incentivesfor good teachers, in comparisonwith only about half of public schools. While both public and private school principals ensured that teaching materials were availablein the school, principals of private schools also ensured that teachingmaterials were readily available to teachers by providingthem storage cabinets in their classrooms. They also protected the school building and its content by ensuring regular maintenanceand by taking precautions against vandalism. In short, they used their decisionmakingpowers to improvethe conditionsfor learning in the schools.


29

Table12: TheDecisionsover whichthe Principalhas the most Influencein Publicand PrivateSchoolsin the Minisample, Colombia,DominicanRepublic,Philippines,Tanzaniaand Thailand, 1990 (Percentof Schools) Area Selectingteachers Selectingnonteacherstaff Dismissingschoolpersonnel Selectingteachersfor in-service Evaluatingteacherperformance Adaptingthe curriculum Establishingstandardsfor student promotion Improvinginstructionalpractice Choosingtextbooks Purchasingequipment Establishinghomeworkpolicies Selectingstudents Selectingschoolfees Using schoolfees

Private

Public

41.9 51.6 45.2 63.0 54.8 16.1

27.6 17.2 37.9 40.7 57.1 13.8

16.1 25.8 16.7 58.1 6.5 45.2 29.0 41.9

6.9 17.2 0.0 34.5 10.7 27.6 14.8 34.6

One reason that private school principals may have paid more attention to teaching and learningis that they, themselves,weremore likelyto be involvedin teaching. On average,private school principalsspent 7.2 hours a week teachinga regular class, comparedwith public school principalswho spent on average only 4.8 hours teachingper week. Private school principals also spent significantly fewer hours on fundraising,communicatingwith the head office and performinggeneral administrative duties than did public school principals(17.1 hours and 24.2 hours respectively). Thus, more of their time was availableto attend to mattersmore directlyrelated to teaching.

30


Table 13: Attention to Teaching and Learning in Public and Private Schools in the Mini-sample, Colombia, DominicanRepublic,Philippines,Tanzaniaand Thailand1990 (Percentof schools) Characteristics

Private

Public

33.3 100.0

17.2 45.4

67.9

50.0

9.7

13.8

Staffmeetingdevotedto "specific pedagogicalpractices' Monetaryrewards for good teaching In-servicetraining devotedto 'better' teachingmethods" "Criticalthinlcingand reasoning skills identifiedas most importanteducationalgoal

V. Significancefor Policy The findings from our research have importantimplicationsfor public educationpolicy. Althoughsome gains in efficiencymay be achievedby mimickingthe mixof resources of private schools (such as teacher/studentratiosand teacherqualifications),this is not likely to be enough to equalizethe two systems. A more effective, albeit less transparent, measure would be to adopt the management practicesof private schools,therebymimickingtheir incentivestructure. This school-levelreorganization would not be possible, however, without significantlychangingthe source of their resources and the bureaucraticstructure in which they are embedded. Thesefindingsshouldnot be interpretedas a call to abolishor privatizepublicschools. The findings are preliminary and need to be tested with other data sets in other environments. Also, the marginaldifferencesfound in the studies may not persist if many students movedfrom public to private schools. Still, the studies do offer initial empirical evidenceon an issue that has to date been subject largely to speculation,often in the context of highly emotionaldebates. One immediateimplicationfor policy is that over-restrictiveregulations on private schools (including outright prohibition in some countries) may be suppressingan efficient way to provide education.

31


Another implicationfor policyis that, in somecases, governmentscouldencouragegreater private sector participationin education. It shouldbe stressed,however,that the relative efficiencyof private schools is highly dependenton the institutionalregimeand structureof the incentivesunder whichthey currently operate. Governmentsubsidies, for example, may not necessarily lead to greater efficiency in the educationalsystem. Such subsidieswill be effectiveonly if they are not linked to restrictions on the schools' ability to choose a suitable inputmix and to strive for greater efficiency. The exact nature of those reforms that lead to improvedefficiencyand equity is beyond the concern of the present book. They might involvecontractingfor educationalservices,as is nowbeingdone in the Philippines,or even some form of vouchersystem,as in Chile.Restrictiverules and regulationsintendedto protectconsumers could be modifiedor tax exemptionscouldbe grantedfor private schools. All such measureswill have to be discussedin the larger contextof the politicaleconomyof specificcountries(James 1987). A final implicationfor policy is that public schools could emulateat least some of the teaching and administrativepracticesof their private counterparts. The usual assumptionin consideringgovernment policies toward private schools is that the quality of educationthey provide is not commensuratewith what is being paid by the consumers, due to the asymmetryof informationbetween consumersand providers. This widely held assumptionis complementedby the view that bureaucrats have better informationregardingthe technologyof education.The evidence,however, is that privateschools,which are more autonomousand responsive to students and their parents, will deliver educationin a costeffectiveway. Althoughthe rigorous methodologywe haveused in comparingpublic and private schools has allowedsome clear advancesin the literature,additionalwork is warranted. First, the databaseswere not strictly comparable across countries and it is not possible to make cross-countrygeneralizations. Second, the scope of countriescoveredis also limited. Third, better information,particularlyregarding the social and privatecosts of differentkindsof schools,needsto be gathered. Fourth, it wouldbe useful

32


to compareresults acrossthe entire distributionof studentsrather thanjust for the averagestudent. And finally, the studies covered only secondaryschools. In Latin Americanand East Asia, the critical level for the future will be universities, which are the highest cost componentsin many budgets for public education. In Africa and the Indian subcontinent,the issue is also being discussedat the primary level.

REF1ERENCES AmericanSociologicalAssociation.(1982)Sociologyof Education. 55(2). AmericanSociologicalAssociation.(1985) Sociologyof Education. 58(2). Birdsall, N. and James, E. Efficiency and equity in social spending: How and why governments misbehave. In J. van der Gaag (Ed.), Includingthe Poor. Washington,D.C.: World Bank. Cohen, J. (1969). Statisticalpower analysisfor the behavioralsciences. New York: AcademicPress. Coleman,J. and Hoffer, T. (1987). Public and private high schools:The impactof communities. New York: BasicBooks. Coleman,J., Hoffer, T. and Kilgore, S. (1982). High school achievement:public. catholicand private schools compared. New York: Basic Books. Cox, D. and Jimenez, E. (1991). The relative effectivenessof private and publicschools:Evidencefrom two developingcountries. Journal of DevelopmentEconomics. 34. 99-121. Hannaway,J. (1991). The organizationand managementof public and catholicschools: Looking inside the "black box". InternationalJournal of EducationalResearch. 15(5),463-481. Hanushek, E. A. (1986). The economicsof schooling:Productionand efficiencyin the public schools. Journal of EconomicLiterature, 25, 1141-77. Hanushek, E. A. (1990). Issues in the public-private split of educational provision. Unpublished manuscript,Universityof Rochester, New York. James, E. (1988).Benefitsand Costs of PrivatizedPublicServices:Lessonsfrom the Dutch Educational System. In GeoffreyWalford (Ed.), Private Schoolsin Ten Countries:Policy and Practice. New York. Routledge and Kegan Paul. James, E. (1989). Publicand PrivateEducationin InternationalPerspective.In W.L. Boydand J. Cibulka (Eds.). Private Schoolsand PublicPolicy: InternationalPerspectives(213-236). London: Falmer Press.

33


James, E. (1989).The Private NonprofitProvisionof Education:A TheoreticalModel and Application to Japan. In E. James (Ed.) The Nonprofit Sectorin InternationalPerspective:Studies in comparative culture and policv. New York: Oxford UniversityPress. James, E. (1989).The private provisionof publicservices:A comparisonof Swedenand Holland.In E. James (Ed.) The NonprofitSectorin InternationalPerspective:Studiesin comparativeculture and policy. New York: Oxford UniversityPress. James, E. (1991).Publicpoliciestowardprivateeducation.InternationalJournalof EducationalResearch. 15 359-376. James, E. and Birdsall, N. (1991) Public versus private provision of social services: Is there an efficiency-equitytrade-off? In K. McCarthy (Ed.), The Nonprofit Sector in the Global Community (pp.51-69). New York: Jossey Bass. James, E. (1987).Publicpoliciestoward private education. (DiscussionPaper, Educationand Training Series, Report No. EDT84). Washington,D.C.: The World Bank. James, E. (1986).Differencesin the role of the private educationalsector in modern and developing countries. Unpublishedmanuscript,State Universityof New York, Stonybrook. Jimenez, E., Lockheed, M., Luna, E. and Paqueo, V. (1991). School effects and costs for private and publicschoolsin the DominicanRepublic.InternationalJournal of EducationalResearch. 15(5),393-410. Jimenez, E., Lockheed,M. and Paqueo,V. (1991). The relativeefficiencyof private and public schools in developingcountries.The WorldBank ResearchObserver,6(2):139-164. Jimenez, E., Lockheed,M. and Wattanawaha,N. (1988). The relative efficiencyof public and private schools: The case of Thailand. The World BankEconomicReview, 2 (2), 139-164. Jimenez, E., Paqueo, V. and Vera, M. L. de (1988). Studentperformanceand school costs in private and public high schools in the Philippines. (PolicyResearchWorkingPaper No. 61). Washington,D.C.: The World Bank. Jimenez, E. and Lockheed, M. (Eds.). (1991). Private Versus Public Education: An International Perspective.InternationalJournal of EducationalResearch 1S(5). Laya,J. and Associates.(1987).Philippinesecondaryeducationfinancingand rate of return. Unpublished manuscript, Bureau of SecondaryEducationand SecondaryEducationDevelopmentBudget, Manila, Philippines. Lee, V. E. and Bryk, A.S. (1989). A multilevelmodel of the social distribution of high school achievement. Sociologyof Education.61 (2), 78-94. Lee, V. E. and Lockheed,M.E. (1990).The effectsof single-sexschoolingon achievementand attitudes in Nigeria. 34(2), 209-231.

34


Levin, H. M. (1987) Education as a public and private good. Journal of Policy Analysis and Management.6(4), 62841. Lockheed, M.E. and Zhao, Q. (1993). The empty opportunity:Local control of secondaryschools and studentachievementin the Philippines.InternationalJournal of EducationDevelopment,13 (1), 45-62. Lockheed, M.E. and Verspoor,A. M. (1991). ImprovingPrimarvEducationin DevelopingCountries. New York: OxfordUniversityPress. Lockheed, M. E., Thorpe, M. , Brooks-Gunn,J., Casserly, P., and McAloon, A. (1985) Sex and ethnic differences in middle school mathematics.science and computer science: What do we know? Report Submittedto the Ford Foundation. Princeton:EducationTesting Service. Lockheed, M.E., Vail, StephenC., Fuller, B. (1986).How textbooksaffect achievementin developing countries:Evidencefrom Thailand.EducationalEvaluationand Policy Analysis. 8(4), 379-92.

35

Lockheedand JimenezPage 35

Luna, E. and Gonzalez,S. (1986). The underdevelopment of mathematicsachievement:Comparisonof public and private schools in the Dominican Regublic. Unpublished manuscript, Centro de InvestigacionesUCMM, DominicanRepublic. Murnane, R. J., Newstead, S. and Olsen, R. (1985). Comparingpublic and private schools: The puzzling role of selectivitybias. Journal of Businessand EconomicStatistics,3(1), 23-35. Murnane, R. J. (1986). Comparisonsof private and public schools: The critical role of regulations. In Daniel C. Levy (Ed.) Private education: Studies in choice and public policy. New York: Oxford UniversityPress. Murnane, R. J. (1985). Comparisonsof public and private schools:Lessonsfrom the uproar. Journal of Human Resources, 20, 263-267.

OECD. (1990). Education in OECD Countries. 1987-88: A Compendiumof StatisticalInformation. OECD: Paris. Psacharopoulos,G. and Loxley,W. (1985). Diversifiedsecondaryeducationand development:Evidence from Colombiaand Tanzania. New York: OxfordUniversityPress. Psacharopoulos, G. (1987). Public versus private schools in developing countries: Evidence from Colombiaand Tanzania. InternationalJournal for EducationalDevelopment. 7(1), 59-67. Robitaille, D. F. and Garden, R. A. (1989). The IEA studyof mathematicsII: Context and outcomes of school mathematics. Oxford: PergamonPress. Roth, G. (1987). Private provision of public services in developing countries. New York: Oxford UniversityPress. Samoff,J. (1991). Local initiativesand nationalpolicies:The politics of private schoolingin Tanzania. InternationalJournal of EducationalResearch. 15(5), 377-391. Samoff, J. (1987) School expansionin Tanzania: Private intiatives and public policy. Comparative EducationReview.31(3), 333-360. Tan, E.A. (1991) Structure and performanceof Phillipinesecondaryschools.Unpublishedmanuscript, World Bank, Washington,D.C. The World Bank (1986). Financingeducationin developingcountries:An explorationof policyoptions. Washington,D.C.: The World Bank. The World Bank (1990). World developmentreport 1990.Poverty, Oxford UniversityPress. Tsang, M. and Taoklam,W. (1990, March).The costs of governmentand private primary educationin Thailand. Paper presented at the ComparativeInternationalEducationSociety Conference, Anaheim, California. Unesco(1991). [UNESCOdata base official enrollmentseries]. Paris: UNESCO.

36


Wattanawaha,N. (1986). A study of eguity in mathematicsteaching and learning in lower secondary schools in Thailand.Unpublisheddoctoral dissertation,Universityof Illinois, Champaign-Urbana.